SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2301.02039 · NeurIPS · 2022

Randomized Message-Interception Smoothing: Gray-box Certificates for Graph Neural Networks

Stephan Günnemann, Aleksandar Bojchevski, Simon Geisler, Jan Schuchardt, Yan Scholten

arXiv · PDF · Open in the Atlas

Code that ran

3 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 2 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2301.02039.json · badge: /api/ran/2301.02039.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
yascho/interception_smoothing community
Papers with Code archive
3 ran · 0 no recorded run
FunctionStatusWhere it lives
compute_bounds Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
yascho/interception_smoothing/interception_smoothing/cert.py
code served (permissive licence) · get_code("2faf5c28fa14d6ea")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Build the all_paths structure
# all_paths[target][source] = list of paths
# Each path is a list of edges
all_paths = {
    0: {
        1: [[(1, 0)], [(1, 0), (2, 0)]],  # Two paths from source 1 to target 0
        2: [[(2, 0)]],  # One path from source 2 to target 0
    },
    1: {
        0: [[(0, 1)], [(0, 1), (3, 1)]],  # Two paths from source 0 to target 1
        3: [[(3, 1)]],  # One path from source 3 to target 1
    },
}

# Build the nodes structure
# nodes[target] = set of source nodes that can reach target
nodes = {
    0: [1, 2],
    1: [0, 3],
}

# Target node indices to compute bounds for
idx = np.array([0, 1])

# Probabilities
p_nodes = 0.1  # Node feature ablation probability
p_edges = 0.2  # Edge deletion probability

ARGS = {
    "all_paths": all_paths,
    "nodes": nodes,
    "idx": idx,
    "p_nodes": p_nodes,
    "p_edges": p_edges,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_delta_single_source Ran
ran, honoured the contract we drafted
yascho/interception_smoothing/interception_smoothing/cert.py
code served (permissive licence) · get_code("bb3352d62f54167c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float64, ndim 0, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Construct paths dict
# The function expects paths to be a dict where keys are something and values are lists of paths
# Looking at the code: len(set([p_edges[0][0] for p_edges in paths])) == 1
# This suggests paths is iterable and each element has p_edges[0][0]
# Actually, looking more carefully: [1 - (1 - p_edges) ** len(p) for p in paths]
# So paths is a list of paths, where each path p has a len()
# And p_edges[0][0] suggests indexing into the dict

# Let's construct a simple case:
# paths should be a dict mapping something to a list of paths
# Each path is a sequence (list or tuple) representing nodes on a path
paths = {
    'source_1': [
        [0, 1, 2],      # path with 3 nodes
        [0, 3]          # path with 2 nodes
    ]
}

# Node feature ablation probability
p_nodes = 0.1

# Edge deletion probability - this should be a scalar based on usage
# The code does (1 - p_edges) ** len(p)
# So p_edges is a single float value
p_edges = 0.2

ARGS = {
    "paths": paths,
    "p_nodes": p_nodes,
    "p_edges": p_edges
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_paths Ran
ran; our contract draft was wrong, not the code
yascho/interception_smoothing/interception_smoothing/cert.py
code served (permissive licence) · get_code("644cb644cd0d6d27")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# Construct the arguments for get_paths
# all_paths: dict mapping target -> dict mapping selected_key -> list of paths
# target: a key in all_paths
# selected: a list of keys to look up in all_paths[target]

all_paths = {
    "target_A": {
        "select_1": ["path_1", "path_2"],
        "select_2": ["path_3"],
        "select_3": ["path_4", "path_5", "path_6"]
    },
    "target_B": {
        "select_1": ["path_x"],
        "select_2": ["path_y", "path_z"]
    }
}

target = "target_A"
selected = ["select_1", "select_3"]

ARGS = {
    "all_paths": all_paths,
    "target": target,
    "selected": selected
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Randomized smoothing is one of the most promising frameworks for certifying the adversarial robustness of machine learning models, including Graph Neural Networks (GNNs). Yet, existing randomized smoothing certificates for GNNs are overly pessimistic since they treat the model as a black box, ignoring the underlying architecture. To remedy this, we propose novel gray-box certificates that exploit the message-passing principle of GNNs: We randomly intercept messages and carefully analyze the probability that messages from adversarially controlled nodes reach their target nodes. Compared to existing certificates, we certify robustness to much stronger adversaries that control entire nodes in the graph and can arbitrarily manipulate node features. Our certificates provide stronger guarantees for attacks at larger distances, as messages from farther-away nodes are more likely to get intercepted. We demonstrate the effectiveness of our method on various models and datasets. Since our gray-box certificates consider the underlying graph structure, we can significantly improve certifiable robustness by applying graph sparsification. 1

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2301.02039")
get_code_for_paper("2301.02039")
have("2301.02039")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2301.02039.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2301.02039.svg)](https://syntology.ai/paper/2301.02039)

Connect an agent — have() is free.